Intelligent factory quality management system based on big data
By designing a smart factory quality management system based on big data in an intelligent factory, using computer vision technology and Internet of Things technology, the problem of difficulty in self-inspection of the quality management system is solved, automated quality management is realized, and production efficiency and system are improved stability.
Patent Information
- Application Number
- CN202411269682.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-11
AI Technical Summary
In intelligent factories, it is difficult to self-inspecify the quality management system, which makes it difficult to balance the data processing efficiency and the timeliness of quality management, which may lead to data backlog and system failure.
Design an intelligent factory quality management system based on big data, including real-time monitoring module, data drive module, closed-loop control module and offset alarm module, and realize automated quality management through computer vision technology and Internet of Things technology.
It realizes automation of intelligent factory quality management, reduces the operating burden of staff, improves the real-time and efficiency of production processes, and avoids data backlog and system failure.
Smart Images

Figure CN118778585B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of Internet of Things, and more specifically, to an intelligent factory quality management system based on big data. Background Art
[0002] An intelligent factory is a factory that improves the production efficiency, flexibility and overall management level of the factory by introducing advanced information technology, automation technology, Internet of Things, big data, artificial intelligence, etc. on the basis of traditional manufacturing. The intelligent factory aims to achieve a highly automated, intelligent and digital production process, so that the factory can self-perceive, self-learn, self-determine and self-optimize. However, in the actual operation of the intelligent factory, it is difficult to evaluate the effectiveness of the factory's own intelligent quality management. The production process involves a large amount of data that needs to be processed urgently. It is difficult to balance the efficiency of data processing and the timeliness of quality management. This may lead to a large amount of data backlog causing downtime, which directly causes the failure of the quality management system.
[0003] In order to solve the above defects, a technical solution is now proposed. Summary of the invention
[0004] The purpose of the present invention is to provide an intelligent factory quality management system based on big data to solve the problems in the above-mentioned background technology.
[0005] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an intelligent factory quality management system based on big data, comprising a real-time monitoring module, a data-driven module, a closed-loop control module, and an offset alarm module;
[0006] The real-time monitoring module is used to monitor the process parameters of the production equipment in real time through sensors and Internet of Things technology, and pre-process the monitoring data before transmitting it to the data driving module;
[0007] The data-driven module is used to use computer vision technology to model and analyze the pre-processed real-time monitoring data;
[0008] The closed-loop control module is used to adjust process parameters according to the modeling analysis results and realize real-time adjustment of the production process through automated control;
[0009] The deviation alarm module is used to provide decision support for quality management based on the process effects of real-time adjustments.
[0010] Preferably, the method of data preprocessing is:
[0011] Data cleaning: Process missing values and outliers in real-time monitoring data. The methods for processing missing values include directly deleting records containing missing values, filling missing values with the median or mean, and using linear interpolation methods to fill missing values. The methods for processing outliers include directly deleting outliers, smoothing data with local weighted regression methods, and replacing outliers with the mean of normal values before and after or interpolation methods.
[0012] Data denoising: Filter the data. The filtering methods include low-pass filtering, high-pass filtering, and band-pass filtering to remove high-frequency noise in the data and use smoothing algorithms to reduce fluctuations in the data.
[0013] Data standardization: Standardization of real-time monitoring data is achieved through Box-Cox transformation.
[0014] Preferably, in the processing of missing values, the specific method of using linear interpolation to fill the missing values is:
[0015] For two known data points and , determine the unknown point between two points The linear interpolation formula used is .
[0016] Preferably, the method of smoothing the data using the local weighted regression method is:
[0017] Define the Gaussian kernel function and calculate the weight of each data point based on the distance. The expression of the Gaussian kernel function is , where is the current forecast point, is a data point, is the smoothing parameter;
[0018] A weighted least squares method is used to fit a local linear model to each prediction point. The fitting method is , where is the target variable, is the independent variable, is the intercept of the model, is the slope of the model, minimizing the weighted square error to obtain and , the method to minimize the weighted square error is , where is the loss function representing the weighted squared error, n is the number of data points, i is the index of the data point, is the weight represented by the Gaussian kernel function, is the true output value of the ith data point, is the input variable value of the ith data point, is the predicted value of the i-th data point using the linear model;
[0019] For each prediction point, a local linear model is used to make predictions and finally smoothed data is obtained.
[0020] Preferably, in data denoising, the method of using a smoothing algorithm to reduce fluctuations in the data is:
[0021] The initial state estimate and the initial state covariance matrix are set based on the initial data. The initial state estimate is used to represent the state of the system at the initial moment, and the initial state covariance matrix is used to represent the uncertainty of the initial state.
[0022] Predict the current state and the state covariance matrix at the current moment. The state prediction equation is: , where is the predicted state estimate at time t, is the state transfer matrix, is the state estimate at time t-1, is the control matrix, is the control input at time t-1, and the prediction equation of the state covariance matrix is , where is the predicted state covariance matrix at time t, is the state covariance matrix at time t-1, is the process noise covariance matrix;
[0023] The state estimate and state covariance matrix are updated respectively by calculating the Kalman gain, and the prediction and update steps are repeated until the residual of the Kalman filter converges.
[0024] Preferably, the method for modeling analysis is:
[0025] Establish a target analysis model and calibrate the target state as , the calculation expression is , where Qv is the defect index, Bu is the behavior index, and Et is the equipment index. are the weight coefficients of defect index, behavior index and equipment index respectively, and are all positive integers;
[0026] The defect index is obtained by inspecting the appearance quality and size specifications of the product. The appearance quality and size specifications of the product are identified through image processing and pattern recognition technology. The size parameters of the product are measured through edge detection and contour extraction methods in the image. When it is detected that the product has surface scratches, dents, uneven colors or size specifications that do not meet the standards, the defect index value is 1, otherwise it is 0;
[0027] The behavior index analyzes the worker's operation through motion recognition technology and verifies whether the worker wears safety equipment correctly. If it is detected that the worker's operation is not standardized or the worker does not wear safety equipment correctly, the behavior index value is 1, otherwise it is 0;
[0028] The equipment index uses pattern recognition technology to analyze the operating status of the equipment, detect whether the equipment is operating normally, and whether there are abnormal conditions such as vibration, heat, or leakage. When it is detected that the equipment fails to operate normally or an abnormal failure occurs, the equipment index takes a value of 1, otherwise it takes a value of 0.
[0029] Preferably, the logic for achieving real-time adjustment of the production process through automated control is:
[0030] Set targets for process parameter adjustments, setting constraints based on equipment physical limitations, safety regulations, and production costs;
[0031] Based on the results of the modeling analysis, an optimization algorithm is used to determine the adjusted parameter settings to meet the optimization goals;
[0032] Through the closed-loop control system, the changes in the production process are monitored in real time, and parameter adjustment suggestions are generated in a timely manner based on the analysis results;
[0033] The corresponding control instructions are automatically generated according to the optimized process parameters. After receiving the control instructions, the control system performs parameter adjustments to change the operating status of the equipment.
[0034] Preferably, the method for decision support is:
[0035] Mark the downtime and rework caused by quality problems, record the downtime as Dot, the number of reworks in the cycle as Nre, and the rework quantity as Req, then calculate the mass balance correction coefficient, the expression is , where Mbc is the mass balance correction coefficient. When the calculated mass balance correction coefficient is greater than or equal to the preset balance threshold Mth, it indicates that the quality management system has a quality inspection fault that cannot be adjusted by itself, prompting maintenance personnel to intervene and repair it. When the calculated mass balance correction coefficient is less than the preset balance threshold Mth, it indicates that the quality inspection quality of the quality management system meets the standard, indicating that the quality management is normal.
[0036] In the above technical solution, the technical effects and advantages provided by the present invention are:
[0037] This application combines computer vision technology with the Internet of Things technology to connect sensor data to achieve automated quality management in smart factories. It uses target recognition and edge detection to simultaneously detect product defects, work operations, and equipment status, conducts a comprehensive analysis and verification of the production process, and evaluates the effectiveness of quality management based on shutdown and rework status. It effectively solves the problem that the quality management system is difficult to self-check the system itself, reduces the operating burden on staff, and improves the production utilization rate of industrial equipment. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0039] Figure 1 It is a system module diagram of the present invention. DETAILED DESCRIPTION
[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0041] Example 1: Please refer to Figure 1 As shown, the present invention is an intelligent factory quality management system based on big data, including a real-time monitoring module, a data driving module, a closed-loop control module, and an offset alarm module;
[0042] The real-time monitoring module is used to monitor the process parameters of the production equipment in real time through sensors and Internet of Things technology, and pre-process the monitoring data before transmitting it to the data driving module;
[0043] The data-driven module is used to use computer vision technology to model and analyze the pre-processed real-time monitoring data;
[0044] The closed-loop control module is used to adjust process parameters according to the modeling analysis results and realize real-time adjustment of the production process through automated control;
[0045] The deviation alarm module is used to provide decision support for quality management based on the process effects of real-time adjustments.
[0046] The real-time monitoring module is used to monitor the process parameters of the production equipment in real time through sensors and Internet of Things technology, and pre-process the monitoring data before transmitting it to the data driving module;
[0047] The sensors include sensors for monitoring the status of production equipment and production process parameters, including temperature sensors for monitoring temperature changes in equipment and production environment to ensure that production is carried out under appropriate temperature conditions;
[0048] Pressure sensors are used to detect pressure in liquids, gases or solids to ensure that the pressure is within the set range to avoid equipment damage or product failure;
[0049] Humidity sensors are used to monitor the humidity of the environment and equipment, especially in humidity-sensitive processes;
[0050] Vibration sensors are used to detect the vibration of equipment and are used for preventive maintenance to avoid equipment failures caused by abnormal vibration;
[0051] Photoelectric sensors are used to detect the position, distance or shape of objects and are often used in automated assembly lines, quality inspection, etc.
[0052] Displacement sensors are used to measure the displacement or movement of objects for precise positioning and monitoring;
[0053] Force sensors are used to monitor the force applied during the production process to ensure that process parameters meet requirements and avoid product quality problems caused by improper force;
[0054] Flow sensors are used to monitor the flow of liquids or gases to ensure a stable supply of materials in the process;
[0055] Current and voltage sensors are used to detect the electrical parameters of equipment and prevent electrical failures;
[0056] Gas sensors are used to detect the concentration of certain specific gases in the air, such as carbon dioxide, oxygen, hydrogen, etc., to ensure the safety of the production environment.
[0057] Through the Internet of Things, various sensors, devices, machines, and control systems are connected together to achieve real-time data collection and transmission. The equipment in the factory is interconnected through wireless or wired networks to form an overall intelligent network.
[0058] The data generated by the equipment is uploaded to the cloud for storage, and cloud computing is used for data analysis and processing to achieve rapid processing of large-scale production data and extract useful information for decision-making.
[0059] The Internet of Things allows factory managers to remotely access the system to monitor and manage equipment. Remote management not only improves efficiency but also reduces labor costs.
[0060] Through the Internet of Things, factories can set up intelligent alarm systems. When certain key equipment or process parameters exceed the preset range, the system will automatically alarm and notify relevant personnel to intervene. In addition, the Internet of Things also supports predictive maintenance, that is, early warning before equipment failure occurs to avoid production stoppage losses.
[0061] The operating status of the equipment is detected by sensors, such as speed, temperature, vibration, etc., to ensure that the equipment works in the best condition. If the equipment is abnormal, the system can automatically adjust or shut down;
[0062] Monitor the energy consumption of equipment in real time, aggregate data through the IoT platform, analyze energy consumption, optimize energy use, and reduce operating costs;
[0063] Record the use of equipment, predict the maintenance and upkeep time of equipment, establish preventive maintenance plans through IoT technology, and reduce equipment failure rates;
[0064] By monitoring the production efficiency of the equipment, analyzing the factors affecting the production efficiency, and improving the efficiency by adjusting the equipment parameters or process flow.
[0065] The process parameters in the real-time monitoring data include:
[0066] Temperature: such as injection molding, heat treatment, food processing, etc.;
[0067] Pressure: In the fields of chemical industry, pharmaceutical industry, hydraulic system, etc., pressure is a key process parameter;
[0068] Flow rate and flow rate: In the process of liquid or gas transportation and mixing, the control of flow rate and flow rate is crucial;
[0069] Time: Processing time, reaction time, etc. need to be precisely controlled;
[0070] Concentration and ratio: In the chemical, food, pharmaceutical and other industries, the concentration and ratio of raw materials are key factors affecting product quality;
[0071] Humidity: Humidity has a great impact on the production process, such as electronics, textiles, and food;
[0072] Physical properties: such as product shape, size, hardness, thickness, etc., may be monitored as process parameters;
[0073] Speed and acceleration: Speed and acceleration are key parameters on high-speed production lines such as printing and packaging.
[0074] By real-time monitoring of these process parameters, the stability and consistency of the production process can be ensured, ultimately improving product quality and reducing production costs.
[0075] Preprocessing the real-time data is an important step to ensure data quality and reliability. In smart factories, data preprocessing can provide a basis for subsequent data analysis, modeling, and decision-making. The methods of data preprocessing are:
[0076] Data cleaning: Process missing values and outliers in real-time monitoring data. The methods for processing missing values include directly deleting records containing missing values, filling missing values with the median or mean, and using linear interpolation methods to fill missing values. The methods for processing outliers include directly deleting outliers, smoothing data with local weighted regression methods, and replacing outliers with the mean of normal values before and after or interpolation methods.
[0077] Data denoising: Filter the data. The filtering methods include low-pass filtering, high-pass filtering, and band-pass filtering to remove high-frequency noise in the data and use smoothing algorithms to reduce fluctuations in the data.
[0078] Data standardization: Standardization of real-time monitoring data is achieved through Box-Cox transformation.
[0079] Real-time monitoring data contains noise. Outliers and missing values in the data will affect the accuracy of the model. Preprocessing can be used to remove noise, correct or delete these abnormal data, ensure data integrity and consistency, make data clearer and more accurate, and thus improve the reliability of analysis and decision-making.
[0080] Real-time data comes from multiple sensors and different process flows. There are inconsistencies between different data sources. Preprocessing is used to standardize and synchronize data so that the data is consistent in time and space. At the same time, by processing missing values and outliers, more data can be effectively utilized and data waste can be reduced. Preprocessing can reduce the dimension and scale of data, reduce computational complexity, and improve the efficiency of subsequent analysis. It is particularly important for real-time monitoring data because the data volume is large and the processing speed requirements are high.
[0081] In the processing of missing values, the specific method of using linear interpolation to fill missing values is:
[0082] For two known data points and , determine the unknown point between two points The linear interpolation formula used is .
[0083] Linear interpolation estimates missing values using the linear relationship between adjacent data points, which can maintain the continuity of data. In many time series or spatial data, data continuity is crucial. Linear interpolation avoids data incoherence by filling missing values smoothly, making the data more coherent and logical when analyzing;
[0084] The calculation of linear interpolation is relatively simple and applicable to various scenarios. Compared with other complex interpolation methods or models, linear interpolation can greatly simplify the processing process while ensuring reasonable accuracy. It is suitable for use in situations where the amount of data is large and computing resources are limited.
[0085] In the process of data analysis, directly deleting records containing missing values may introduce bias, especially when the missing values are not randomly distributed. Linear interpolation estimates missing values by using the trend of existing data points, reducing data bias caused by improper handling of missing values and ensuring the integrity of the data and the reliability of the analysis results;
[0086] For simple linear trend data, using linear interpolation can avoid the additional errors introduced by complex models. Complex models may overfit the data, especially when there are few samples or the data changes are simple, linear interpolation provides a more reasonable way to fill in the gaps;
[0087] In the process of data visualization, missing values may cause breakpoints or discontinuities in the chart. Using linear interpolation to fill missing values can make the chart smoother and easier to understand, avoiding visualization problems caused by missing data and making data trends more intuitive;
[0088] Linear interpolation is suitable for scenarios where data changes linearly locally. In this case, linear interpolation can more accurately reflect the real changes in the data, and the filled data is more consistent with the actual trend. For example, in some physical measurement and environmental monitoring scenarios, data often show linear changes, and linear interpolation is a reasonable choice.
[0089] The method of smoothing data using the local weighted regression method is:
[0090] Define the Gaussian kernel function and calculate the weight of each data point based on the distance. The expression of the Gaussian kernel function is , where is the current forecast point, is a data point, is the smoothing parameter;
[0091] A weighted least squares method is used to fit a local linear model to each prediction point. The fitting method is , where is the target variable, is the independent variable, is the intercept of the model, is the slope of the model, minimizing the weighted square error to obtain and , the method to minimize the weighted square error is , where is the loss function representing the weighted squared error, n is the number of data points, i is the index of the data point, is the weight represented by the Gaussian kernel function, is the true output value of the ith data point, is the input variable value of the ith data point, is the predicted value of the i-th data point using the linear model;
[0092] For each prediction point, a local linear model is used to make predictions and finally smoothed data is obtained.
[0093] Local weighted regression is a non-parametric regression method that does not make any assumptions about the overall structure of the data. Compared with global linear regression, local weighted regression can better capture the local nonlinear characteristics in the data. In practical applications, data often has nonlinear trends. Local weighted regression can fit different linear models in different areas to more accurately reflect the real changes in the data.
[0094] Real data usually contains noise, which interferes with the understanding of data trends. Local weighted regression effectively smooths the data, removes noise, and retains the main trends of the data by assigning weights to each data point. Data points with lower weights have less impact on the fitting results.
[0095] The local weighted regression method can flexibly adapt to local changes in data. In areas where data changes dramatically, local weighted regression can capture changes well; in areas where data is stable, local weighted regression can effectively smooth data. This flexibility makes local weighted regression very effective in dealing with heterogeneous data and is suitable for data sets where different parts show different trends;
[0096] For non-stationary data, where the data trends are inconsistent over time or space, local weighted regression has a particular advantage. Global models are usually affected by extreme values or large-scale trend changes, while local weighted regression can better capture the changing characteristics of data in different time or space ranges by fitting locally;
[0097] Compared with global smoothing methods, local weighted regression can better preserve the local characteristics of the data. While smoothing the data, local weighted regression will not overly smooth the details of the data, and is more suitable for application scenarios where local fluctuations or specific patterns are important.
[0098] In some data sets, the fluctuation range of data points may vary over time or space. This phenomenon is called heteroscedasticity. Local weighted regression effectively reduces the impact of this heteroscedasticity by weighting data points, thereby obtaining a more stable fitting result.
[0099] In data denoising, the method of using smoothing algorithm to reduce fluctuations in data is:
[0100] The initial state estimate and the initial state covariance matrix are set based on the initial data. The initial state estimate is used to represent the state of the system at the initial moment, and the initial state covariance matrix is used to represent the uncertainty of the initial state.
[0101] Predict the current state and the state covariance matrix at the current moment. The state prediction equation is: , where is the predicted state estimate at time t, is the state transfer matrix, is the state estimate at time t-1, is the control matrix, is the control input at time t-1, and the prediction equation of the state covariance matrix is , where is the predicted state covariance matrix at time t, is the state covariance matrix at time t-1, is the process noise covariance matrix;
[0102] The state estimate and state covariance matrix are updated respectively by calculating the Kalman gain, and the prediction and update steps are repeated until the residual of the Kalman filter converges.
[0103] The data-driven module is used to use computer vision technology to model and analyze the pre-processed real-time monitoring data;
[0104] Computer vision technology includes the use of target detection algorithms to detect target objects in images or videos. The target objects include products and equipment parts, and are used to detect whether there are abnormal objects or missing workpieces on the production line;
[0105] Use target recognition technology to classify the target objects to be detected, identify the type of product, the status of the equipment, and the actions of the workers, and use instance segmentation technology to segment each object in the image at the pixel level to achieve detailed analysis;
[0106] Commonly used target detection algorithms include YOLO, Faster R-CNN, etc. Commonly used target recognition technologies include convolutional neural networks CNN, ResNet, etc., and instance segmentation technologies such as Mask R-CNN;
[0107] Establish a target analysis model and calibrate the target state as , the calculation expression is , where Qv is the defect index, Bu is the behavior index, and Et is the equipment index. are the weight coefficients of defect index, behavior index and equipment index respectively, and are all positive integers;
[0108] The defect index is obtained by inspecting the appearance quality and size specifications of the product. The appearance quality and size specifications of the product are identified through image processing and pattern recognition technology. The size parameters of the product are measured through edge detection and contour extraction methods in the image. When it is detected that the product has surface scratches, dents, uneven colors or size specifications that do not meet the standards, the defect index value is 1, otherwise it is 0;
[0109] The behavior index analyzes the worker's operation through motion recognition technology and verifies whether the worker wears safety equipment correctly. If it is detected that the worker's operation is not standardized or the worker does not wear safety equipment correctly, the behavior index value is 1, otherwise it is 0;
[0110] The equipment index uses pattern recognition technology to analyze the operating status of the equipment, detect whether the equipment is operating normally, and whether there are abnormal conditions such as vibration, heat, or leakage. When it is detected that the equipment fails to operate normally or an abnormal failure occurs, the equipment index takes a value of 1, otherwise it takes a value of 0.
[0111] The closed-loop control module is used to adjust process parameters according to the modeling analysis results and realize real-time adjustment of the production process through automated control;
[0112] Set targets for process parameter adjustments, setting constraints based on equipment physical limitations, safety regulations, and production costs;
[0113] Based on the results of the modeling analysis, an optimization algorithm is used to determine the adjusted parameter settings to meet the optimization goals;
[0114] Through the closed-loop control system, the changes in the production process are monitored in real time, and parameter adjustment suggestions are generated in a timely manner based on the analysis results;
[0115] The corresponding control instructions are automatically generated according to the optimized process parameters. After receiving the control instructions, the control system performs parameter adjustments to change the operating status of the equipment.
[0116] The goals of process parameter adjustment include minimizing product defect rates, maximizing production efficiency, and reducing energy consumption. The adjustment goals are related to quality, efficiency, and production safety indicators. Constraints such as equipment temperature and pressure cannot exceed the safety range. Optimization algorithms include genetic algorithms and particle swarm algorithms.
[0117] The deviation alarm module is used to support quality management decisions based on the process effects of real-time adjustments;
[0118] Mark the downtime and rework caused by quality problems, record the downtime as Dot, the number of reworks in the cycle as Nre, and the rework quantity as Req, then calculate the mass balance correction coefficient, the expression is , where Mbc is the mass balance correction coefficient. When the calculated mass balance correction coefficient is greater than or equal to the preset balance threshold Mth, it indicates that the quality management system has a quality inspection fault that cannot be adjusted by itself, prompting maintenance personnel to intervene and repair it. When the calculated mass balance correction coefficient is less than the preset balance threshold Mth, it indicates that the quality inspection quality of the quality management system meets the standard, indicating that the quality management is normal.
[0119] This application combines computer vision technology with the Internet of Things technology to connect sensor data to achieve automated quality management in smart factories. It uses target recognition and edge detection to simultaneously detect product defects, work operations, and equipment status, conducts a comprehensive analysis and verification of the production process, and evaluates the effectiveness of quality management based on shutdown and rework status. It effectively solves the problem that the quality management system is difficult to self-check the system itself, reduces the operating burden on staff, and improves the production utilization rate of industrial equipment.
[0120] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.
[0121] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.
[0122] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0123] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the method described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.
Claims
1. Intelligent factory quality management system based on big data, characterized by: It includes real-time monitoring module, data driving module, closed-loop control module and deviation alarm module; The real-time monitoring module is used to monitor the process parameters of the production equipment in real time through sensors and Internet of Things technology, and pre-process the monitoring data before transmitting it to the data driving module; The data-driven module is used to use computer vision technology to model and analyze the pre-processed real-time monitoring data; The closed-loop control module is used to adjust process parameters according to the modeling analysis results and realize real-time adjustment of the production process through automated control; The deviation alarm module is used to support quality management decisions based on the process effects of real-time adjustments; The method for modeling analysis is: Establish a target analysis model and calibrate the target state as , the calculation expression is , where Qv is the defect index, Bu is the behavior index, and Et is the equipment index. are the weight coefficients of defect index, behavior index and equipment index respectively, and All are positive integers; The defect index is obtained by inspecting the appearance quality and size specifications of the product. The appearance quality and size specifications of the product are identified through image processing and pattern recognition technology. The size parameters of the product are measured through edge detection and contour extraction methods in the image. When it is detected that the product has surface scratches, dents, uneven colors or size specifications that do not meet the standards, the defect index value is 1, otherwise it is 0; The behavior index analyzes the worker's operation through motion recognition technology and verifies whether the worker wears safety equipment correctly. If it is detected that the worker's operation is not standardized or the worker does not wear safety equipment correctly, the behavior index value is 1, otherwise it is 0; The equipment index uses pattern recognition technology to analyze the operating status of the equipment, detect whether the equipment is operating normally, and whether there are abnormal conditions such as vibration, heating, and leakage. When it is detected that the equipment fails to operate normally or an abnormal failure occurs, the equipment index takes a value of 1, otherwise it takes a value of 0; The logic of achieving real-time adjustment of production processes through automated control is: Set targets for process parameter adjustments, setting constraints based on equipment physical limitations, safety regulations, and production costs; Based on the results of the modeling analysis, an optimization algorithm is used to determine the adjusted parameter settings to meet the optimization goals; Through the closed-loop control system, the changes in the production process are monitored in real time, and parameter adjustment suggestions are generated in a timely manner based on the analysis results; Automatically generate corresponding control instructions based on optimized process parameters. After receiving the control instructions, the control system will adjust the parameters and change the operating status of the equipment. The decision support methods are: Mark the downtime and rework caused by quality problems, record the downtime as Dot, the number of reworks in the cycle as Nre, and the rework quantity as Req, then calculate the mass balance correction coefficient, the expression is , where Mbc is the mass balance correction coefficient. When the calculated mass balance correction coefficient is greater than or equal to the preset balance threshold Mth, it indicates that the quality management system has a quality inspection fault that cannot be adjusted by itself, prompting maintenance personnel to intervene and repair it. When the calculated mass balance correction coefficient is less than the preset balance threshold Mth, it indicates that the quality inspection quality of the quality management system meets the standard, indicating that the quality management is normal.
2. The intelligent factory quality management system based on big data according to claim 1 is characterized in that: The data preprocessing method is: Data cleaning: Process missing values and outliers in real-time monitoring data. The methods for processing missing values include directly deleting records containing missing values, filling missing values with the median or mean, and using linear interpolation methods to fill missing values. The methods for processing outliers include directly deleting outliers, smoothing data with local weighted regression methods, and replacing outliers with the mean of normal values before and after or interpolation methods. Data denoising: Filter the data. The filtering methods include low-pass filtering, high-pass filtering, and band-pass filtering to remove high-frequency noise in the data and use smoothing algorithms to reduce fluctuations in the data. Data standardization: Standardization of real-time monitoring data is achieved through Box-Cox transformation.
3. The intelligent factory quality management system based on big data according to claim 2 is characterized in that: In the processing of missing values, the specific method of using linear interpolation to fill missing values is: For two known data points and , determine the unknown point between two points The linear interpolation formula used is .
4. The intelligent factory quality management system based on big data according to claim 2 is characterized in that: The method of smoothing data using the local weighted regression method is: Define the Gaussian kernel function and calculate the weight of each data point based on the distance. The expression of the Gaussian kernel function is , where is the current forecast point, is a data point, is the smoothing parameter; A weighted least squares method is used to fit a local linear model to each prediction point. The fitting method is , where is the target variable, is the independent variable, is the intercept of the model, is the slope of the model, minimizing the weighted square error to obtain and , the method to minimize the weighted square error is , where is the loss function representing the weighted squared error, n is the number of data points, i is the index of the data point, is the weight represented by the Gaussian kernel function, is the true output value of the ith data point, is the input variable value of the ith data point, is the predicted value of the i-th data point using the linear model; For each prediction point, a local linear model is used to make predictions and finally smoothed data is obtained.
5. The intelligent factory quality management system based on big data according to claim 2 is characterized in that: In data denoising, the method of using smoothing algorithm to reduce fluctuations in data is: The initial state estimate and the initial state covariance matrix are set based on the initial data. The initial state estimate is used to represent the state of the system at the initial moment, and the initial state covariance matrix is used to represent the uncertainty of the initial state. Predict the current state and the state covariance matrix at the current moment. The state prediction equation is: , where is the predicted state estimate at time t, is the state transfer matrix, is the state estimate at time t-1, is the control matrix, is the control input at time t-1, and the prediction equation of the state covariance matrix is , where is the predicted state covariance matrix at time t, is the state covariance matrix at time t-1, is the process noise covariance matrix; The state estimate and state covariance matrix are updated respectively by calculating the Kalman gain, and the prediction and update steps are repeated until the residual of the Kalman filter converges.
Citation Information
Patent Citations
Intelligent factory production online monitoring and analysis system based on machine vision
CN116841262A
Data mining analysis method
CN117453764A
